
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
%matplotlib inline

import plotly.express as px
import plotly.graph_objects as go
from plotly.subplots import make_subplots

import warnings
warnings.simplefilter(action="ignore", category=FutureWarning)
warnings.simplefilter(action="ignore", category=Warning)


df = pd.read_csv(r'titanic.csv')
df.head()


df.dtypes

pclass       float64
survived     float64
name          object
sex           object
age          float64
sibsp        float64
parch        float64
ticket        object
fare         float64
cabin         object
embarked      object
boat          object
body         float64
home.dest     object
dtype: object


df.isnull().sum()

pclass          1
survived        1
name            1
sex             1
age           264
sibsp           1
parch           1
ticket          1
fare            2
cabin        1015
embarked        3
boat          824
body         1189
home.dest     565
dtype: int64


df['age'].fillna(df['age'].mean(),inplace=True)
df.head(10)


for i, col in enumerate(['sibsp', 'parch']):
    plt.figure(i)
    sns.catplot(x=col, y = 'survived', data = df, kind = 'point', aspect=2, )

<Figure size 432x288 with 0 Axes>


df['family_size'] = df['sibsp'] +  df['parch']


df.drop(['sibsp', 'parch', 'body', 'boat'], axis=1, inplace=True)
df.head()


df.isnull().sum()

pclass            1
survived          1
name              1
sex               1
age               0
ticket            1
fare              2
cabin          1015
embarked          3
home.dest       565
family_size       1
dtype: int64


df.groupby(df['cabin'].isnull())['survived'].mean()

cabin
False    0.654237
True     0.302761
Name: survived, dtype: float64


df['has_cabin'] = np.where(df['cabin'].isnull(), 0, 1)
df.head(30)


gender_map = {'male': 0, 'female': 1}
df['sex'] = df['sex'].map(gender_map)
df.head()


df.groupby(df['embarked'].isnull())['survived'].mean()

embarked
False    0.381025
True     1.000000
Name: survived, dtype: float64


df.groupby(df['home.dest'].isnull())['survived'].mean()

home.dest
False    0.465772
True     0.271277
Name: survived, dtype: float64


df.drop(['cabin', 'embarked', 'name', 'ticket', 'home.dest'], axis=1, inplace=True)
df.head()


df.isnull().sum()

pclass         1
survived       1
sex            1
age            0
fare           2
family_size    1
has_cabin      0
dtype: int64


df.ffill(axis = 0, inplace=True)


df.to_csv('cleaned_titanic.csv')


df.describe

<bound method NDFrame.describe of       pclass  survived  sex        age      fare  family_size  has_cabin
0        1.0       1.0  1.0  29.000000  211.3375          0.0          1
1        1.0       1.0  0.0   0.916700  151.5500          3.0          1
2        1.0       0.0  1.0   2.000000  151.5500          3.0          1
3        1.0       0.0  0.0  30.000000  151.5500          3.0          1
4        1.0       0.0  1.0  25.000000  151.5500          3.0          1
...      ...       ...  ...        ...       ...          ...        ...
1305     3.0       0.0  1.0  29.881135   14.4542          1.0          0
1306     3.0       0.0  0.0  26.500000    7.2250          0.0          0
1307     3.0       0.0  0.0  27.000000    7.2250          0.0          0
1308     3.0       0.0  0.0  29.000000    7.8750          0.0          0
1309     3.0       0.0  0.0  29.881135    7.8750          0.0          0

[1310 rows x 7 columns]>


features=df.columns


fig = make_subplots(rows=1, cols=2)
fig.add_trace(go.Box( y=df[features[1]]),row=1,col=1)
fig.add_trace(go.Histogram(x=df[features[1]]),row=1,col=2)
fig.update_layout(height=600, title_text="Boxplot and histogram for feature: "+str(features[1]))
fig.update_layout(showlegend=False)
fig.show()


fig = make_subplots(rows=1, cols=2)
fig.add_trace(go.Box( y=df[features[2]]),row=1,col=1)
fig.add_trace(go.Histogram(x=df[features[2]]),row=1,col=2)
fig.update_layout(height=600, title_text="Boxplot and histogram for feature: "+str(features[2]))
fig.update_layout(showlegend=False)
fig.show()


fig = make_subplots(rows=1, cols=2)
fig.add_trace(go.Box( y=df[features[3]]),row=1,col=1)
fig.add_trace(go.Histogram(x=df[features[3]]),row=1,col=2)
fig.update_layout(height=600, title_text="Boxplot and histogram for feature: "+str(features[3]))
fig.update_layout(showlegend=False)
fig.show()


fig = make_subplots(rows=1, cols=2)
fig.add_trace(go.Box( y=df[features[4]]),row=1,col=1)
fig.add_trace(go.Histogram(x=df[features[4]]),row=1,col=2)
fig.update_layout(height=600, title_text="Boxplot and histogram for feature: "+str(features[4]))
fig.update_layout(showlegend=False)
fig.show()


fig = make_subplots(rows=1, cols=2)
fig.add_trace(go.Box( y=df[features[5]]),row=1,col=1)
fig.add_trace(go.Histogram(x=df[features[5]]),row=1,col=2)
fig.update_layout(height=600, title_text="Boxplot and histogram for feature: "+str(features[5]))
fig.update_layout(showlegend=False)
fig.show()


fig = make_subplots(rows=1, cols=2)
fig.add_trace(go.Box( y=df[features[6]]),row=1,col=1)
fig.add_trace(go.Histogram(x=df[features[6]]),row=1,col=2)
fig.update_layout(height=600, title_text="Boxplot and histogram for feature: "+str(features[6]))
fig.update_layout(showlegend=False)
fig.show()


from sklearn.model_selection import train_test_split
df = pd.read_csv(r'cleaned_titanic.csv')
df.head()


features = df.drop('survived', axis=1)
target = df['survived']

X_train, X_test, y_train, y_test = train_test_split(features, target, test_size=0.4, random_state = 1965)
X_val, X_test, y_val, y_test = train_test_split(X_test, y_test, test_size=0.5, random_state = 1965)


for dataset in [y_train, y_val, y_test]:
    print(round(len(dataset) / len(target), 2))

0.6
0.2
0.2


X_train.to_csv('train_features.csv', index = False)
X_val.to_csv('val_features.csv', index = False)
X_test.to_csv('test_features.csv', index = False)

y_train.to_csv('train_target.csv', index=False)
y_val.to_csv('val_target.csv', index=False)
y_test.to_csv('test_target.csv', index=False)


from sklearn.ensemble import GradientBoostingClassifier, AdaBoostClassifier
import joblib
from sklearn.model_selection import GridSearchCV

#Examine all the accessible hyperparameters 
GradientBoostingClassifier().get_params()

{'ccp_alpha': 0.0,
 'criterion': 'friedman_mse',
 'init': None,
 'learning_rate': 0.1,
 'loss': 'deviance',
 'max_depth': 3,
 'max_features': None,
 'max_leaf_nodes': None,
 'min_impurity_decrease': 0.0,
 'min_impurity_split': None,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'n_estimators': 100,
 'n_iter_no_change': None,
 'presort': 'deprecated',
 'random_state': None,
 'subsample': 1.0,
 'tol': 0.0001,
 'validation_fraction': 0.1,
 'verbose': 0,
 'warm_start': False}


#Examine all the accessible hyperparameters 

AdaBoostClassifier().get_params()

{'algorithm': 'SAMME.R',
 'base_estimator': None,
 'learning_rate': 1.0,
 'n_estimators': 50,
 'random_state': None}


tr_features = pd.read_csv(r'train_features.csv')
tr_target = pd.read_csv(r'train_target.csv')


tr_features.shape

(786, 7)


tr_target.shape

(786, 1)


tr_target.isnull().sum()

survived    0
dtype: int64


tr_features.isnull().sum()

Unnamed: 0     0
pclass         0
sex            0
age            0
fare           0
family_size    0
has_cabin      0
dtype: int64


def print_results(results):
    print('The best hyperparameter are: {}\n'.format(results.best_params_))
    
    means = results.cv_results_['mean_test_score']
    stds = results.cv_results_['std_test_score']
    for mean, std, params in zip(means, stds, results.cv_results_['params']):
        print('{} (+/-{}) for {}'.format(round(mean, 3), round(std * 2, 3), params))


gb = GradientBoostingClassifier()
parameters = {
    'n_estimators': [5, 50, 250, 500],
    'max_depth': [1, 3, 5, 7, 9],
    'learning_rate': [0.01, 0.1, 1, 10, 100]
}
cv = GridSearchCV(gb, parameters, cv=5)
cv.fit(tr_features, tr_target.values.ravel()) # ravel changes column to an array

print_results(cv)

The best hyperparameter are: {'learning_rate': 0.01, 'max_depth': 7, 'n_estimators': 250}

0.627 (+/-0.006) for {'learning_rate': 0.01, 'max_depth': 1, 'n_estimators': 5}
0.726 (+/-0.113) for {'learning_rate': 0.01, 'max_depth': 1, 'n_estimators': 50}
0.766 (+/-0.073) for {'learning_rate': 0.01, 'max_depth': 1, 'n_estimators': 250}
0.783 (+/-0.06) for {'learning_rate': 0.01, 'max_depth': 1, 'n_estimators': 500}
0.627 (+/-0.006) for {'learning_rate': 0.01, 'max_depth': 3, 'n_estimators': 5}
0.791 (+/-0.035) for {'learning_rate': 0.01, 'max_depth': 3, 'n_estimators': 50}
0.813 (+/-0.023) for {'learning_rate': 0.01, 'max_depth': 3, 'n_estimators': 250}
0.816 (+/-0.015) for {'learning_rate': 0.01, 'max_depth': 3, 'n_estimators': 500}
0.627 (+/-0.006) for {'learning_rate': 0.01, 'max_depth': 5, 'n_estimators': 5}
0.789 (+/-0.039) for {'learning_rate': 0.01, 'max_depth': 5, 'n_estimators': 50}
0.817 (+/-0.028) for {'learning_rate': 0.01, 'max_depth': 5, 'n_estimators': 250}
0.819 (+/-0.035) for {'learning_rate': 0.01, 'max_depth': 5, 'n_estimators': 500}
0.627 (+/-0.006) for {'learning_rate': 0.01, 'max_depth': 7, 'n_estimators': 5}
0.808 (+/-0.025) for {'learning_rate': 0.01, 'max_depth': 7, 'n_estimators': 50}
0.822 (+/-0.019) for {'learning_rate': 0.01, 'max_depth': 7, 'n_estimators': 250}
0.816 (+/-0.008) for {'learning_rate': 0.01, 'max_depth': 7, 'n_estimators': 500}
0.627 (+/-0.006) for {'learning_rate': 0.01, 'max_depth': 9, 'n_estimators': 5}
0.796 (+/-0.056) for {'learning_rate': 0.01, 'max_depth': 9, 'n_estimators': 50}
0.809 (+/-0.017) for {'learning_rate': 0.01, 'max_depth': 9, 'n_estimators': 250}
0.817 (+/-0.025) for {'learning_rate': 0.01, 'max_depth': 9, 'n_estimators': 500}
0.726 (+/-0.113) for {'learning_rate': 0.1, 'max_depth': 1, 'n_estimators': 5}
0.784 (+/-0.049) for {'learning_rate': 0.1, 'max_depth': 1, 'n_estimators': 50}
0.786 (+/-0.04) for {'learning_rate': 0.1, 'max_depth': 1, 'n_estimators': 250}
0.79 (+/-0.059) for {'learning_rate': 0.1, 'max_depth': 1, 'n_estimators': 500}
0.791 (+/-0.035) for {'learning_rate': 0.1, 'max_depth': 3, 'n_estimators': 5}
0.812 (+/-0.022) for {'learning_rate': 0.1, 'max_depth': 3, 'n_estimators': 50}
0.807 (+/-0.031) for {'learning_rate': 0.1, 'max_depth': 3, 'n_estimators': 250}
0.799 (+/-0.062) for {'learning_rate': 0.1, 'max_depth': 3, 'n_estimators': 500}
0.788 (+/-0.041) for {'learning_rate': 0.1, 'max_depth': 5, 'n_estimators': 5}
0.817 (+/-0.032) for {'learning_rate': 0.1, 'max_depth': 5, 'n_estimators': 50}
0.804 (+/-0.035) for {'learning_rate': 0.1, 'max_depth': 5, 'n_estimators': 250}
0.804 (+/-0.021) for {'learning_rate': 0.1, 'max_depth': 5, 'n_estimators': 500}
0.808 (+/-0.025) for {'learning_rate': 0.1, 'max_depth': 7, 'n_estimators': 5}
0.812 (+/-0.026) for {'learning_rate': 0.1, 'max_depth': 7, 'n_estimators': 50}
0.805 (+/-0.043) for {'learning_rate': 0.1, 'max_depth': 7, 'n_estimators': 250}
0.799 (+/-0.039) for {'learning_rate': 0.1, 'max_depth': 7, 'n_estimators': 500}
0.796 (+/-0.053) for {'learning_rate': 0.1, 'max_depth': 9, 'n_estimators': 5}
0.805 (+/-0.024) for {'learning_rate': 0.1, 'max_depth': 9, 'n_estimators': 50}
0.789 (+/-0.042) for {'learning_rate': 0.1, 'max_depth': 9, 'n_estimators': 250}
0.799 (+/-0.03) for {'learning_rate': 0.1, 'max_depth': 9, 'n_estimators': 500}
0.788 (+/-0.07) for {'learning_rate': 1, 'max_depth': 1, 'n_estimators': 5}
0.788 (+/-0.033) for {'learning_rate': 1, 'max_depth': 1, 'n_estimators': 50}
0.785 (+/-0.035) for {'learning_rate': 1, 'max_depth': 1, 'n_estimators': 250}
0.78 (+/-0.026) for {'learning_rate': 1, 'max_depth': 1, 'n_estimators': 500}
0.793 (+/-0.03) for {'learning_rate': 1, 'max_depth': 3, 'n_estimators': 5}
0.77 (+/-0.023) for {'learning_rate': 1, 'max_depth': 3, 'n_estimators': 50}
0.779 (+/-0.019) for {'learning_rate': 1, 'max_depth': 3, 'n_estimators': 250}
0.789 (+/-0.047) for {'learning_rate': 1, 'max_depth': 3, 'n_estimators': 500}
0.771 (+/-0.039) for {'learning_rate': 1, 'max_depth': 5, 'n_estimators': 5}
0.786 (+/-0.013) for {'learning_rate': 1, 'max_depth': 5, 'n_estimators': 50}
0.8 (+/-0.036) for {'learning_rate': 1, 'max_depth': 5, 'n_estimators': 250}
0.795 (+/-0.028) for {'learning_rate': 1, 'max_depth': 5, 'n_estimators': 500}
0.791 (+/-0.026) for {'learning_rate': 1, 'max_depth': 7, 'n_estimators': 5}
0.796 (+/-0.039) for {'learning_rate': 1, 'max_depth': 7, 'n_estimators': 50}
0.803 (+/-0.023) for {'learning_rate': 1, 'max_depth': 7, 'n_estimators': 250}
0.768 (+/-0.03) for {'learning_rate': 1, 'max_depth': 7, 'n_estimators': 500}
0.78 (+/-0.053) for {'learning_rate': 1, 'max_depth': 9, 'n_estimators': 5}
0.804 (+/-0.036) for {'learning_rate': 1, 'max_depth': 9, 'n_estimators': 50}
0.782 (+/-0.044) for {'learning_rate': 1, 'max_depth': 9, 'n_estimators': 250}
0.78 (+/-0.035) for {'learning_rate': 1, 'max_depth': 9, 'n_estimators': 500}
0.234 (+/-0.073) for {'learning_rate': 10, 'max_depth': 1, 'n_estimators': 5}
0.234 (+/-0.073) for {'learning_rate': 10, 'max_depth': 1, 'n_estimators': 50}
0.234 (+/-0.073) for {'learning_rate': 10, 'max_depth': 1, 'n_estimators': 250}
0.234 (+/-0.073) for {'learning_rate': 10, 'max_depth': 1, 'n_estimators': 500}
0.332 (+/-0.301) for {'learning_rate': 10, 'max_depth': 3, 'n_estimators': 5}
0.325 (+/-0.31) for {'learning_rate': 10, 'max_depth': 3, 'n_estimators': 50}
0.325 (+/-0.31) for {'learning_rate': 10, 'max_depth': 3, 'n_estimators': 250}
0.325 (+/-0.31) for {'learning_rate': 10, 'max_depth': 3, 'n_estimators': 500}
0.444 (+/-0.12) for {'learning_rate': 10, 'max_depth': 5, 'n_estimators': 5}
0.441 (+/-0.114) for {'learning_rate': 10, 'max_depth': 5, 'n_estimators': 50}
0.445 (+/-0.115) for {'learning_rate': 10, 'max_depth': 5, 'n_estimators': 250}
0.443 (+/-0.117) for {'learning_rate': 10, 'max_depth': 5, 'n_estimators': 500}
0.59 (+/-0.098) for {'learning_rate': 10, 'max_depth': 7, 'n_estimators': 5}
0.543 (+/-0.136) for {'learning_rate': 10, 'max_depth': 7, 'n_estimators': 50}
0.582 (+/-0.109) for {'learning_rate': 10, 'max_depth': 7, 'n_estimators': 250}
0.547 (+/-0.137) for {'learning_rate': 10, 'max_depth': 7, 'n_estimators': 500}
0.635 (+/-0.104) for {'learning_rate': 10, 'max_depth': 9, 'n_estimators': 5}
0.629 (+/-0.109) for {'learning_rate': 10, 'max_depth': 9, 'n_estimators': 50}
0.63 (+/-0.105) for {'learning_rate': 10, 'max_depth': 9, 'n_estimators': 250}
0.617 (+/-0.12) for {'learning_rate': 10, 'max_depth': 9, 'n_estimators': 500}
0.373 (+/-0.006) for {'learning_rate': 100, 'max_depth': 1, 'n_estimators': 5}
0.373 (+/-0.006) for {'learning_rate': 100, 'max_depth': 1, 'n_estimators': 50}
0.373 (+/-0.006) for {'learning_rate': 100, 'max_depth': 1, 'n_estimators': 250}
0.373 (+/-0.006) for {'learning_rate': 100, 'max_depth': 1, 'n_estimators': 500}
0.282 (+/-0.056) for {'learning_rate': 100, 'max_depth': 3, 'n_estimators': 5}
0.275 (+/-0.033) for {'learning_rate': 100, 'max_depth': 3, 'n_estimators': 50}
0.282 (+/-0.056) for {'learning_rate': 100, 'max_depth': 3, 'n_estimators': 250}
0.275 (+/-0.033) for {'learning_rate': 100, 'max_depth': 3, 'n_estimators': 500}
0.419 (+/-0.122) for {'learning_rate': 100, 'max_depth': 5, 'n_estimators': 5}
0.427 (+/-0.125) for {'learning_rate': 100, 'max_depth': 5, 'n_estimators': 50}
0.419 (+/-0.123) for {'learning_rate': 100, 'max_depth': 5, 'n_estimators': 250}
0.42 (+/-0.122) for {'learning_rate': 100, 'max_depth': 5, 'n_estimators': 500}
0.482 (+/-0.067) for {'learning_rate': 100, 'max_depth': 7, 'n_estimators': 5}
0.424 (+/-0.075) for {'learning_rate': 100, 'max_depth': 7, 'n_estimators': 50}
0.421 (+/-0.089) for {'learning_rate': 100, 'max_depth': 7, 'n_estimators': 250}
0.452 (+/-0.046) for {'learning_rate': 100, 'max_depth': 7, 'n_estimators': 500}
0.555 (+/-0.141) for {'learning_rate': 100, 'max_depth': 9, 'n_estimators': 5}
0.608 (+/-0.079) for {'learning_rate': 100, 'max_depth': 9, 'n_estimators': 50}
0.534 (+/-0.143) for {'learning_rate': 100, 'max_depth': 9, 'n_estimators': 250}
0.564 (+/-0.118) for {'learning_rate': 100, 'max_depth': 9, 'n_estimators': 500}


cv.best_estimator_

GradientBoostingClassifier(learning_rate=0.01, max_depth=7, n_estimators=250)


joblib.dump(cv.best_estimator_, 'model_gb.pkl')

['model_gb.pkl']


from sklearn.ensemble import RandomForestClassifier 

#Examine all the accessible hyperparameters 
RandomForestClassifier().get_params()

{'bootstrap': True,
 'ccp_alpha': 0.0,
 'class_weight': None,
 'criterion': 'gini',
 'max_depth': None,
 'max_features': 'auto',
 'max_leaf_nodes': None,
 'max_samples': None,
 'min_impurity_decrease': 0.0,
 'min_impurity_split': None,
 'min_samples_leaf': 1,
 'min_samples_split': 2,
 'min_weight_fraction_leaf': 0.0,
 'n_estimators': 100,
 'n_jobs': None,
 'oob_score': False,
 'random_state': None,
 'verbose': 0,
 'warm_start': False}


def print_results(results):
    print('The best hyperparameter are: {}\n'.format(results.best_params_))
    
    means = results.cv_results_['mean_test_score']
    stds = results.cv_results_['std_test_score']
    for mean, std, params in zip(means, stds, results.cv_results_['params']):
        print('{} (+/-{}) for {}'.format(round(mean, 3), round(std * 2, 3), params))


rf = RandomForestClassifier()
parameters = {
    'n_estimators': [5, 50, 250, 500, 1000],
    'max_depth': [4, 8, 16, 32, None]
}
cv = GridSearchCV(rf, parameters, cv=5)
cv.fit(tr_features, tr_target.values.ravel()) # ravel changes column vector to array

print_results(cv)

The best hyperparameter are: {'max_depth': 8, 'n_estimators': 250}

0.784 (+/-0.074) for {'max_depth': 4, 'n_estimators': 5}
0.794 (+/-0.057) for {'max_depth': 4, 'n_estimators': 50}
0.798 (+/-0.049) for {'max_depth': 4, 'n_estimators': 250}
0.803 (+/-0.049) for {'max_depth': 4, 'n_estimators': 500}
0.802 (+/-0.051) for {'max_depth': 4, 'n_estimators': 1000}
0.793 (+/-0.067) for {'max_depth': 8, 'n_estimators': 5}
0.812 (+/-0.052) for {'max_depth': 8, 'n_estimators': 50}
0.818 (+/-0.034) for {'max_depth': 8, 'n_estimators': 250}
0.817 (+/-0.037) for {'max_depth': 8, 'n_estimators': 500}
0.814 (+/-0.031) for {'max_depth': 8, 'n_estimators': 1000}
0.802 (+/-0.061) for {'max_depth': 16, 'n_estimators': 5}
0.807 (+/-0.05) for {'max_depth': 16, 'n_estimators': 50}
0.81 (+/-0.049) for {'max_depth': 16, 'n_estimators': 250}
0.81 (+/-0.034) for {'max_depth': 16, 'n_estimators': 500}
0.807 (+/-0.045) for {'max_depth': 16, 'n_estimators': 1000}
0.782 (+/-0.059) for {'max_depth': 32, 'n_estimators': 5}
0.802 (+/-0.055) for {'max_depth': 32, 'n_estimators': 50}
0.817 (+/-0.033) for {'max_depth': 32, 'n_estimators': 250}
0.803 (+/-0.041) for {'max_depth': 32, 'n_estimators': 500}
0.808 (+/-0.048) for {'max_depth': 32, 'n_estimators': 1000}
0.763 (+/-0.081) for {'max_depth': None, 'n_estimators': 5}
0.807 (+/-0.055) for {'max_depth': None, 'n_estimators': 50}
0.812 (+/-0.043) for {'max_depth': None, 'n_estimators': 250}
0.807 (+/-0.049) for {'max_depth': None, 'n_estimators': 500}
0.807 (+/-0.043) for {'max_depth': None, 'n_estimators': 1000}


cv.best_estimator_

RandomForestClassifier(max_depth=8, n_estimators=250)


joblib.dump(cv.best_estimator_, 'model_rf.pkl')

['model_rf.pkl']


from sklearn.ensemble import StackingClassifier


#Examine all the accessible hyperparameters 
estimators = [('gb',GradientBoostingClassifier()), ('rf', RandomForestClassifier())]
StackingClassifier(estimators = estimators).get_params()

{'cv': None,
 'estimators': [('gb', GradientBoostingClassifier()),
  ('rf', RandomForestClassifier())],
 'final_estimator': None,
 'n_jobs': None,
 'passthrough': False,
 'stack_method': 'auto',
 'verbose': 0,
 'gb': GradientBoostingClassifier(),
 'rf': RandomForestClassifier(),
 'gb__ccp_alpha': 0.0,
 'gb__criterion': 'friedman_mse',
 'gb__init': None,
 'gb__learning_rate': 0.1,
 'gb__loss': 'deviance',
 'gb__max_depth': 3,
 'gb__max_features': None,
 'gb__max_leaf_nodes': None,
 'gb__min_impurity_decrease': 0.0,
 'gb__min_impurity_split': None,
 'gb__min_samples_leaf': 1,
 'gb__min_samples_split': 2,
 'gb__min_weight_fraction_leaf': 0.0,
 'gb__n_estimators': 100,
 'gb__n_iter_no_change': None,
 'gb__presort': 'deprecated',
 'gb__random_state': None,
 'gb__subsample': 1.0,
 'gb__tol': 0.0001,
 'gb__validation_fraction': 0.1,
 'gb__verbose': 0,
 'gb__warm_start': False,
 'rf__bootstrap': True,
 'rf__ccp_alpha': 0.0,
 'rf__class_weight': None,
 'rf__criterion': 'gini',
 'rf__max_depth': None,
 'rf__max_features': 'auto',
 'rf__max_leaf_nodes': None,
 'rf__max_samples': None,
 'rf__min_impurity_decrease': 0.0,
 'rf__min_impurity_split': None,
 'rf__min_samples_leaf': 1,
 'rf__min_samples_split': 2,
 'rf__min_weight_fraction_leaf': 0.0,
 'rf__n_estimators': 100,
 'rf__n_jobs': None,
 'rf__oob_score': False,
 'rf__random_state': None,
 'rf__verbose': 0,
 'rf__warm_start': False}


from sklearn.linear_model import LogisticRegression


def print_results(results):
    print('The optimal hyper parameters for this model are: {}\n'.format(results.best_params_))
    
    means = results.cv_results_['mean_test_score']
    stds = results.cv_results_['std_test_score']
    for mean, std, params in zip(means, stds, results.cv_results_['params']):
        print('{} (+/-{}) for {}'.format(round(mean, 3), round(std * 2, 3), params))


estimators = [('rf', RandomForestClassifier()),
             ('gb', GradientBoostingClassifier())]

sc = StackingClassifier(estimators=estimators)
sc.get_params()

{'cv': None,
 'estimators': [('rf', RandomForestClassifier()),
  ('gb', GradientBoostingClassifier())],
 'final_estimator': None,
 'n_jobs': None,
 'passthrough': False,
 'stack_method': 'auto',
 'verbose': 0,
 'rf': RandomForestClassifier(),
 'gb': GradientBoostingClassifier(),
 'rf__bootstrap': True,
 'rf__ccp_alpha': 0.0,
 'rf__class_weight': None,
 'rf__criterion': 'gini',
 'rf__max_depth': None,
 'rf__max_features': 'auto',
 'rf__max_leaf_nodes': None,
 'rf__max_samples': None,
 'rf__min_impurity_decrease': 0.0,
 'rf__min_impurity_split': None,
 'rf__min_samples_leaf': 1,
 'rf__min_samples_split': 2,
 'rf__min_weight_fraction_leaf': 0.0,
 'rf__n_estimators': 100,
 'rf__n_jobs': None,
 'rf__oob_score': False,
 'rf__random_state': None,
 'rf__verbose': 0,
 'rf__warm_start': False,
 'gb__ccp_alpha': 0.0,
 'gb__criterion': 'friedman_mse',
 'gb__init': None,
 'gb__learning_rate': 0.1,
 'gb__loss': 'deviance',
 'gb__max_depth': 3,
 'gb__max_features': None,
 'gb__max_leaf_nodes': None,
 'gb__min_impurity_decrease': 0.0,
 'gb__min_impurity_split': None,
 'gb__min_samples_leaf': 1,
 'gb__min_samples_split': 2,
 'gb__min_weight_fraction_leaf': 0.0,
 'gb__n_estimators': 100,
 'gb__n_iter_no_change': None,
 'gb__presort': 'deprecated',
 'gb__random_state': None,
 'gb__subsample': 1.0,
 'gb__tol': 0.0001,
 'gb__validation_fraction': 0.1,
 'gb__verbose': 0,
 'gb__warm_start': False}


parameters = {
    'gb__n_estimators': [50, 100, 250],
    'rf__n_estimators': [50, 100, 250],
    'final_estimator': [LogisticRegression(C=0.1),
                        LogisticRegression(C=1),
                        LogisticRegression(C=10),
                        LogisticRegression(C=50),],
    'passthrough': [True, False]
}
cv = GridSearchCV(sc, parameters, cv=5)
cv.fit(tr_features, tr_target.values.ravel())

print_results(cv)

The optimal hyper parameters for this model are: {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 250}

0.804 (+/-0.046) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 50}
0.804 (+/-0.038) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 100}
0.8 (+/-0.046) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 250}
0.819 (+/-0.03) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 50}
0.816 (+/-0.023) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 100}
0.816 (+/-0.029) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 250}
0.805 (+/-0.047) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 50}
0.799 (+/-0.04) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 100}
0.804 (+/-0.041) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 250}
0.823 (+/-0.039) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 50}
0.826 (+/-0.039) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 100}
0.817 (+/-0.034) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 250}
0.81 (+/-0.036) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 50}
0.802 (+/-0.036) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 100}
0.807 (+/-0.037) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 250}
0.814 (+/-0.04) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 50}
0.819 (+/-0.049) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 100}
0.823 (+/-0.036) for {'final_estimator': LogisticRegression(C=0.1), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 250}
0.8 (+/-0.034) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 50}
0.809 (+/-0.036) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 100}
0.803 (+/-0.037) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 250}
0.819 (+/-0.025) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 50}
0.817 (+/-0.022) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 100}
0.818 (+/-0.013) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 250}
0.812 (+/-0.036) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 50}
0.804 (+/-0.046) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 100}
0.803 (+/-0.043) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 250}
0.826 (+/-0.027) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 50}
0.824 (+/-0.021) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 100}
0.826 (+/-0.023) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 250}
0.804 (+/-0.04) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 50}
0.807 (+/-0.038) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 100}
0.805 (+/-0.04) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 250}
0.817 (+/-0.038) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 50}
0.817 (+/-0.041) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 100}
0.817 (+/-0.037) for {'final_estimator': LogisticRegression(C=1), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 250}
0.803 (+/-0.036) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 50}
0.807 (+/-0.041) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 100}
0.805 (+/-0.048) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 250}
0.814 (+/-0.033) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 50}
0.812 (+/-0.029) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 100}
0.813 (+/-0.026) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 250}
0.805 (+/-0.034) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 50}
0.803 (+/-0.044) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 100}
0.81 (+/-0.045) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 250}
0.821 (+/-0.03) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 50}
0.819 (+/-0.033) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 100}
0.817 (+/-0.032) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 250}
0.813 (+/-0.032) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 50}
0.809 (+/-0.029) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 100}
0.807 (+/-0.04) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 250}
0.817 (+/-0.036) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 50}
0.819 (+/-0.031) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 100}
0.814 (+/-0.025) for {'final_estimator': LogisticRegression(C=10), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 250}
0.804 (+/-0.044) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 50}
0.802 (+/-0.04) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 100}
0.805 (+/-0.04) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 50, 'passthrough': True, 'rf__n_estimators': 250}
0.808 (+/-0.029) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 50}
0.812 (+/-0.029) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 100}
0.814 (+/-0.03) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 50, 'passthrough': False, 'rf__n_estimators': 250}
0.804 (+/-0.04) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 50}
0.805 (+/-0.04) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 100}
0.804 (+/-0.034) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 100, 'passthrough': True, 'rf__n_estimators': 250}
0.818 (+/-0.034) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 50}
0.819 (+/-0.03) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 100}
0.817 (+/-0.028) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 100, 'passthrough': False, 'rf__n_estimators': 250}
0.807 (+/-0.034) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 50}
0.81 (+/-0.025) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 100}
0.809 (+/-0.024) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 250, 'passthrough': True, 'rf__n_estimators': 250}
0.817 (+/-0.042) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 50}
0.813 (+/-0.032) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 100}
0.813 (+/-0.032) for {'final_estimator': LogisticRegression(C=50), 'gb__n_estimators': 250, 'passthrough': False, 'rf__n_estimators': 250}


joblib.dump(cv.best_estimator_, 'stacked_mod.pkl')

['stacked_mod.pkl']


# Import standard performance libraries
from sklearn.metrics import accuracy_score, precision_score, recall_score 
from time import time


BG_Model = joblib.load('model_gb.pkl')
RF_Model = joblib.load('model_rf.pkl')
STACKED_Model = joblib.load('stacked_mod.pkl')

val_features = pd.read_csv('val_features.csv')
val_target = pd.read_csv('val_target.csv')


def evaluate_model(model, features, target):
    start = time()
    pred = model.predict(features)
    end = time()
    accuracy = round(accuracy_score(target, pred), 3)
    precision = round(precision_score(target, pred), 3)
    recall = round(recall_score(target, pred), 3)
    print('{} -- Accuracy: {} / Precision: {} / Recall: {} / Latency: {}ms'.format(str(model).split('(')[0],
                                                                                   accuracy,
                                                                                   precision,
                                                                                   recall,
                                                                                   round((end - start)*1000, 1)))


for mdl in [BG_Model, RF_Model, STACKED_Model]:
    evaluate_model(mdl, val_features, val_target)

GradientBoostingClassifier -- Accuracy: 0.798 / Precision: 0.802 / Recall: 0.676 / Latency: 3.0ms
RandomForestClassifier -- Accuracy: 0.779 / Precision: 0.784 / Recall: 0.639 / Latency: 21.0ms
StackingClassifier -- Accuracy: 0.786 / Precision: 0.783 / Recall: 0.667 / Latency: 24.0ms


test_features = pd.read_csv('test_features.csv')
test_target = pd.read_csv('test_target.csv')

evaluate_model(BG_Model, test_features, test_target)

GradientBoostingClassifier -- Accuracy: 0.84 / Precision: 0.843 / Recall: 0.707 / Latency: 3.0ms

The Titanic¶

Survivor Predictions¶

Table of Contents

Goals and Objectives¶

Project Overview¶

Import Dependancies¶

Read in and view dataset¶

Data Cleaning - Continuous Features¶

Data Cleaning - Categorical Features¶

Visualize Cleaned Dataset¶

Partition dataset¶

* Train, test, and validation split¶

* Validate partition size.¶

* Save partitioned datasets¶

Ensemble 'Boosting' Model¶

Hyperparameter Tuning¶

Save boosting model¶

Ensemble 'Bagging' Model¶

Hyperparameter Tuning¶

Save bagging model¶

Ensemble 'Stacking' Model¶

Save stacked model¶

Comparing the different ensemble models¶

Validation Set Evaluation¶

Assessing the best performing model (gradient-boosting) on the test data.¶

Conclusion:¶

	pclass	survived	name	sex	age	sibsp	parch	ticket	fare	cabin	embarked	boat	body	home.dest
0	1.0	1.0	Allen, Miss. Elisabeth Walton	female	29.0000	0.0	0.0	24160	211.3375	B5	S	2	NaN	St Louis, MO
1	1.0	1.0	Allison, Master. Hudson Trevor	male	0.9167	1.0	2.0	113781	151.5500	C22 C26	S	11	NaN	Montreal, PQ / Chesterville, ON
2	1.0	0.0	Allison, Miss. Helen Loraine	female	2.0000	1.0	2.0	113781	151.5500	C22 C26	S	NaN	NaN	Montreal, PQ / Chesterville, ON
3	1.0	0.0	Allison, Mr. Hudson Joshua Creighton	male	30.0000	1.0	2.0	113781	151.5500	C22 C26	S	NaN	135.0	Montreal, PQ / Chesterville, ON
4	1.0	0.0	Allison, Mrs. Hudson J C (Bessie Waldo Daniels)	female	25.0000	1.0	2.0	113781	151.5500	C22 C26	S	NaN	NaN	Montreal, PQ / Chesterville, ON

	pclass	survived	name	sex	age	ticket	fare	cabin	embarked	home.dest	family_size	has_cabin
0	1.0	1.0	Allen, Miss. Elisabeth Walton	female	29.000000	24160	211.3375	B5	S	St Louis, MO	0.0	1
1	1.0	1.0	Allison, Master. Hudson Trevor	male	0.916700	113781	151.5500	C22 C26	S	Montreal, PQ / Chesterville, ON	3.0	1
2	1.0	0.0	Allison, Miss. Helen Loraine	female	2.000000	113781	151.5500	C22 C26	S	Montreal, PQ / Chesterville, ON	3.0	1
3	1.0	0.0	Allison, Mr. Hudson Joshua Creighton	male	30.000000	113781	151.5500	C22 C26	S	Montreal, PQ / Chesterville, ON	3.0	1
4	1.0	0.0	Allison, Mrs. Hudson J C (Bessie Waldo Daniels)	female	25.000000	113781	151.5500	C22 C26	S	Montreal, PQ / Chesterville, ON	3.0	1
5	1.0	1.0	Anderson, Mr. Harry	male	48.000000	19952	26.5500	E12	S	New York, NY	0.0	1
6	1.0	1.0	Andrews, Miss. Kornelia Theodosia	female	63.000000	13502	77.9583	D7	S	Hudson, NY	1.0	1
7	1.0	0.0	Andrews, Mr. Thomas Jr	male	39.000000	112050	0.0000	A36	S	Belfast, NI	0.0	1
8	1.0	1.0	Appleton, Mrs. Edward Dale (Charlotte Lamson)	female	53.000000	11769	51.4792	C101	S	Bayside, Queens, NY	2.0	1
9	1.0	0.0	Artagaveytia, Mr. Ramon	male	71.000000	PC 17609	49.5042	NaN	C	Montevideo, Uruguay	0.0	0
10	1.0	0.0	Astor, Col. John Jacob	male	47.000000	PC 17757	227.5250	C62 C64	C	New York, NY	1.0	1
11	1.0	1.0	Astor, Mrs. John Jacob (Madeleine Talmadge Force)	female	18.000000	PC 17757	227.5250	C62 C64	C	New York, NY	1.0	1
12	1.0	1.0	Aubart, Mme. Leontine Pauline	female	24.000000	PC 17477	69.3000	B35	C	Paris, France	0.0	1
13	1.0	1.0	Barber, Miss. Ellen "Nellie"	female	26.000000	19877	78.8500	NaN	S	NaN	0.0	0
14	1.0	1.0	Barkworth, Mr. Algernon Henry Wilson	male	80.000000	27042	30.0000	A23	S	Hessle, Yorks	0.0	1
15	1.0	0.0	Baumann, Mr. John D	male	29.881135	PC 17318	25.9250	NaN	S	New York, NY	0.0	0
16	1.0	0.0	Baxter, Mr. Quigg Edmond	male	24.000000	PC 17558	247.5208	B58 B60	C	Montreal, PQ	1.0	1
17	1.0	1.0	Baxter, Mrs. James (Helene DeLaudeniere Chaput)	female	50.000000	PC 17558	247.5208	B58 B60	C	Montreal, PQ	1.0	1
18	1.0	1.0	Bazzani, Miss. Albina	female	32.000000	11813	76.2917	D15	C	NaN	0.0	1
19	1.0	0.0	Beattie, Mr. Thomson	male	36.000000	13050	75.2417	C6	C	Winnipeg, MN	0.0	1
20	1.0	1.0	Beckwith, Mr. Richard Leonard	male	37.000000	11751	52.5542	D35	S	New York, NY	2.0	1
21	1.0	1.0	Beckwith, Mrs. Richard Leonard (Sallie Monypeny)	female	47.000000	11751	52.5542	D35	S	New York, NY	2.0	1
22	1.0	1.0	Behr, Mr. Karl Howell	male	26.000000	111369	30.0000	C148	C	New York, NY	0.0	1
23	1.0	1.0	Bidois, Miss. Rosalie	female	42.000000	PC 17757	227.5250	NaN	C	NaN	0.0	0
24	1.0	1.0	Bird, Miss. Ellen	female	29.000000	PC 17483	221.7792	C97	S	NaN	0.0	1
25	1.0	0.0	Birnbaum, Mr. Jakob	male	25.000000	13905	26.0000	NaN	C	San Francisco, CA	0.0	0
26	1.0	1.0	Bishop, Mr. Dickinson H	male	25.000000	11967	91.0792	B49	C	Dowagiac, MI	1.0	1
27	1.0	1.0	Bishop, Mrs. Dickinson H (Helen Walton)	female	19.000000	11967	91.0792	B49	C	Dowagiac, MI	1.0	1
28	1.0	1.0	Bissette, Miss. Amelia	female	35.000000	PC 17760	135.6333	C99	S	NaN	0.0	1
29	1.0	1.0	Bjornstrom-Steffansson, Mr. Mauritz Hakan	male	28.000000	110564	26.5500	C52	S	Stockholm, Sweden / Washington, DC	0.0	1